Refactor Validator::decimal() to be more flexible with decimal separators and PHP magic. Lots of testing added.

[bar]

From:

b75a6b4#commitcomment-1767646

[markstory]

Couldn't this just be (?:[0-9]*[\.]{$lnum}) I don't understand why the alternation is needed.

[bar]

Yes, I forgot to remove the grouping, it was needed to make '10.' valid (but with * instead of + at the end).
$dnum = "(?:[0-9]*[\.]{$lnum}|{$lnum}[\.][0-9]+)";
should have been:
$dnum = "[0-9]*[\.]{$lnum}";
I'll amend it.

[bar]

I keep thinking it should be $dnum = "(?:[0-9]*[\.]{$lnum}|{$lnum}[\.][0-9]*)";
so that this is true

$this->assertTrue(Validation::decimal('1234.', null));
$this->assertTrue(Validation::decimal('1234.', true));

because this is true

$this->assertTrue(Validation::decimal(1234., null));
$this->assertTrue(Validation::decimal(1234., true));

[dereuromark]

you dont have to change all the test cases and manually add "null", since the default value is null, you dont have to provide this second argument.

[bar]

I know, but considering the function heavily depends on that parameter, that's a way of making it known. Also, if the default value changes in the (near) future, the test won't break.

[dereuromark]

just saying :)

but besides that. why would you want to use decimal (with null) to validate float? there is an own validation rule "numeric" for that. you would only want to use "decimal" if you actually required the digits after the point - hence this validation rule. isnt your changed decimal pretty equal the numeric one now for null? the same thing does not have to be provided twice I think.

[bar]

BTW, thanks for the hint!

[bar]

I think exactly the same :P

I invite you to read 28bd688 which got itself to master, and later b75a6b4 which fully explains the rationale.

[bar]

Also, take a look at this comment
b75a6b4#commitcomment-1773097

and the outdated diff from this very same PR

[dereuromark]

quite confusing ;) thx anyway

[bar]

It's a mess! sorry, b75a6b4 is an extract